Error processing some command line arguments

Error processing some command line arguments

[Juanma Barranquero]

I'm starting Emacs (HEAD, compiled on W2K) with

  emacs -fn fontset-default

and I get a starting buffer named "fontset-default" (but the argument is
processed). Same happens with

  emacs -fg red

where I get a red foreground *and* a buffer called "red".

Long options (--font=fontset-default, --foreground-color=red) work as
expected.



                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Eli Zaretskii]

> From: Juanma Barranquero <lektu@terra.es>
> Date: Thu, 06 Jun 2002 15:56:24 +0200
> 
> I'm starting Emacs (HEAD, compiled on W2K) with
> 
>   emacs -fn fontset-default
> 
> and I get a starting buffer named "fontset-default" (but the argument is
> processed). Same happens with
> 
>   emacs -fg red
> 
> where I get a red foreground *and* a buffer called "red".

Sounds like Windows-specific: I tried the same on Debian and in the
MS-DOS port, and didn't see the problem.

Does "-q --no-site-file" change anything?

Re: Error processing some command line arguments

[Juanma Barranquero]

On Thu, 06 Jun 2002 18:02:20 +0300, "Eli Zaretskii" <eliz@is.elta.co.il> wrote:

> Sounds like Windows-specific: I tried the same on Debian and in the
> MS-DOS port, and didn't see the problem.

Yes, I think so.

> Does "-q --no-site-file" change anything?

No. In fact I did all tests with -q --no-site-file, and then without
them but without HKLM\SOFTWARE\Gnu\Emacs entries, .emacs nor
site-lisp.el files.

Debugging the issue I saw that in the call to sort_args (in emacs.c),
near the end, the arg is consumed till the following line is executed:

  bcopy (new, argv, sizeof (char *) * argc);

And then it "reappears". But I'm not sure how to debug that.


                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Eli Zaretskii]

> Date: Thu, 06 Jun 2002 18:33:09 +0200
> From: Juanma Barranquero <lektu@terra.es>
> 
> Debugging the issue I saw that in the call to sort_args (in emacs.c),
> near the end, the arg is consumed till the following line is executed:
> 
>   bcopy (new, argv, sizeof (char *) * argc);
> 
> And then it "reappears". But I'm not sure how to debug that.

Does it help to replace `bcopy' with `memmove' (after a suitable
change in the argument list)?

Re: Error processing some command line arguments

[Juanma Barranquero]

On Fri, 07 Jun 2002 16:30:07 +0300, "Eli Zaretskii" <eliz@is.elta.co.il> wrote:

> Does it help to replace `bcopy' with `memmove' (after a suitable
> change in the argument list)?

I tested it and there was no change, but I admit my test weren't very
thourough.


                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Richard Stallman]

    near the end, the arg is consumed till the following line is executed:

      bcopy (new, argv, sizeof (char *) * argc);

"The arg is consumed" could mean many different things.
Could you state what you observe in terms of values of specific
variables and contents of specific data?

Re: Error processing some command line arguments

[Juanma Barranquero]

On Fri, 7 Jun 2002 17:22:45 -0600 (MDT), Richard Stallman <rms@gnu.org> wrote:

> "The arg is consumed" could mean many different things.

Yeah, sorry for being lazy.

> Could you state what you observe in terms of values of specific
> variables and contents of specific data?

I'll do as soon as I can bootstrap Emacs again on W2K. Currently
temacs.exe is crashing while loading international/character.el and I
haven't the foggiest idea how to debug that.


                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Richard Stallman]

    I'll do as soon as I can bootstrap Emacs again on W2K. Currently
    temacs.exe is crashing while loading international/character.el and I
    haven't the foggiest idea how to debug that.

How about running temacs.exe under a debugger and making a backtrace?
That should be a good start.

Re: Error processing some command line arguments

[Juanma Barranquero]

On Tue, 11 Jun 2002 20:32:16 -0600 (MDT), Richard Stallman <rms@gnu.org> wrote:

> How about running temacs.exe under a debugger and making a backtrace?
> That should be a good start.

Er... I already did that and posted a bug report on this list.

Relevant information:



The crash happens at composite.c, line 460:

 func = COMPOSITE_MODIFICATION_FUNC (prop);

The trace stack shows:

 run_composition_function(1, 39987, 1)
 update_compositions(39987, 1, 3)
 Finsert_file_contents("c:/bin/emacs/HEAD/lisp/international/characters.el",
                       nil, 0, 47876, nil)



                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Juanma Barranquero]

On Wed, 12 Jun 2002 08:44:25 +0200, Juanma Barranquero <lektu@terra.es> wrote:

> The crash happens at composite.c, line 460:

It bootstraps fine now, but I don't know what prompted the change : (

Sorry for the noise.

                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Richard Stallman]

    The crash happens at composite.c, line 460:

     func = COMPOSITE_MODIFICATION_FUNC (prop);

    The trace stack shows:

     run_composition_function(1, 39987, 1)
     update_compositions(39987, 1, 3)
     Finsert_file_contents("c:/bin/emacs/HEAD/lisp/international/characters.el",
			   nil, 0, 47876, nil)

Can you figure out what invalid data was the immediate cause of the
crash?

Re: Error processing some command line arguments

[Juanma Barranquero]

On Wed, 12 Jun 2002 17:47:55 -0600 (MDT), Richard Stallman <rms@gnu.org> wrote:

> Can you figure out what invalid data was the immediate cause of the
> crash?

No. I did debug it around for a while and didn't find anything unusual.
Then, after a cvs update and a bootstrap, the problem disappeared.

I'm pretty sure it wasn't any leftover on my system, because I deleted
the entire HEAD source tree, re-checked it out and the problem was still
here. But now it's no more :(


                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Juanma Barranquero]

On Fri, 7 Jun 2002 17:22:45 -0600 (MDT), Richard Stallman <rms@gnu.org> wrote:

> "The arg is consumed" could mean many different things.
> Could you state what you observe in terms of values of specific
> variables and contents of specific data?

My mistake, it was not a C-level bug, but a misplaced parenthesis in
x-handle-switch (term/w32-win.el) so it didn't update x-invocation-args
after processing a "-switch value" pair. It is fixed now.

                                                           /L/e/k/t/u

Re: Error processing some command line arguments

[Richard Stallman]

Thanks.